Clinical event sequences consist of thousands of clinical events that\nrepresent records of patient care in time. Developing accurate prediction\nmodels for such sequences is of a great importance for defining representations\nof a patient state and for improving patient care. One important challenge of\nlearning a good predictive model of clinical sequences is patient-specific\nvariability. Based on underlying clinical complications, each patient's\nsequence may consist of different sets of clinical events. However,\npopulation-based models learned from such sequences may not accurately predict\npatient-specific dynamics of event sequences. To address the problem, we\ndevelop a new adaptive event sequence prediction framework that learns to\nadjust its prediction for individual patients through an online model update.\n